Non-versioned provides with a provider priority as used before this
commit are only useful for virtual packages. That is, packages which
can be installed concurrently to each other. Therefore, having non-version
provider signals that the forked package is intended to be allowed to be
installed alongside the main package. Not only this is not the case (we don't
wanted forked alsa-ucm-conf and the upstream one installed together), but
in addition, doing so is impossible. That is because the main alsa-ucm-conf
package provides by default a versioned provider of itself, making the
concurrent installation not allowed. Therefore, using non-versioned provides,
although harmless in practice, represents a packaging bug that makes
understaning and reasoning about the fork harder.
Instead of doing that, make the providers versioned, and remove the
provider_priority. Now, it is clear that the forked alsa-ucm-conf packages
must replace the upstream one. To avoid installations that do not need a
forked alsa-ucm-conf, we must make sure that the version which the forked
packages provide is lower than the version of the upstream package. We do
so by prepending a 0. to the version, and adding an explanatory comment.
Signed-off-by: Pablo Correa Gómez <pabloyoyoista@postmarketos.org>
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/9397>
zstd is faster to decompress than gzip and will soon
become the default compression algorithm for the initfs,
so might as well take advantage of it early.
The default compression is used because it results in
practically the same binary size as gzip (both create
initramfs binaries that are 30M and zstd:fast adds
a megabyte, which should be avoided). This change is
mostly for the decompression speed benefits.
Signed-off-by: Aster Boese <asterboese@mailbox.org>
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/9191>
- use amove instead of installing to subpackages from main package
- for selecting pixman renderer usage in wlroots, users testing
wlroots0.19-based UIs can just install 0.20 manually to pull in the
configuration, which will also work on 0.19.
Signed-off-By: Sicelo A. Mhlongo <absicsz@gmail.com>
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/9159>
Remove Grub and replace with mainline u-boot & systemd-boot. The two
methods to boot now:
- Hold the RE button to boot off of the SD Card
- With a blank SPI, boot off of eMMC using the power button
SPI needs to be blank as u-boot chainloading is broken
Signed-off-by: Rudraksha Gupta <guptarud@gmail.com>
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/9135>
If your enchilada turns off because the battery is drained and you plug
it in and walk away you will come back to a phone that is still empty
(at least with FDE). With these modules running in the initramfs
charging works. I verified this by pressing the Volume-Down while
booting and in the debug shell I typed:
$ cat /sys/class/power_supply/bq27411-0/current_now
... which returns a negative number while unplugged and a number close
to but below 1500000 if plugged in.
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/9201>
Add more modules to initfs, so that it can detect rootfs on nvme
drive, use usb keyboard and get display properly initialized during
initramfs stage.
Withouth those panel/display is not properly initialized, root partition
cannot be found if it is on nvme drive, and on top of that USB keyboard
is not working, so you can't even debug it from initramfs.
Add myself as co-maintainer of package.
Signed-off-by: Alexey Minnekhanov <alexeymin@postmarketos.org>
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/9203>
The reason why we have patches instead of a separate repo are because of
the following reasons:
1. We don't want to make a repo to which users contribute to. While
people haven't tried to commit to my repository yet, this is
yet another incentive for people to upstream into Linux kernel
mainline. Most of the functionality is here. We just need to
upstream it.
2. I assume that all the patches here aren't needed. This calls
for a scream test. People who find functionalality missing will
open an issue. It becomes trivial to add/remove patches from
the `source=...` var rather than repackaging the entire kernel.
The next commit aims to remove a lot of patches and see if there
are people who rely on said functionality. If not, there is no
requirement to upstream it.
Signed-off-by: Rudraksha Gupta <guptarud@gmail.com>
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/9136>
Currently, loading the a630 firmware files fails in the iniramfs
when the display is manually configured. This firmware is loaded
once the rootfs is loaded and the gpu is polled, but the error
messages clog the early boot logs and can make parsing them much
harder. Because of this, and because these devices don't have an
issue with bootimg size, including the firmware is a QOL change.
Signed-off-by: Aster Boese <asterboese@mailbox.org>
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/8649>
Currently, loading the a630 firmware files fails in the iniramfs
when the display is manually configured. This firmware is loaded
once the rootfs is loaded and the gpu is polled, but the error
messages clog the early boot logs and can make parsing them much
harder. Because of this, and because these devices don't have an
issue with bootimg size, including the firmware is a QOL change.
Signed-off-by: Aster Boese <asterboese@mailbox.org>
Part-of: <https://gitlab.postmarketos.org/postmarketOS/pmaports/-/merge_requests/8649>
Achill Gilgenast